Today's Semiconductors Pulse
AMD CEO says AI boom revives focus on CPUs
AMD CEO Lisa Su said the rapid expansion of AI workloads is refocusing attention on central processing units. AMD also announced it will invest more than $10 billion in Taiwan’s ecosystem to accelerate AI infrastructure deployment.
Sony, TSMC Agree to Cooperate on Next-Generation Image Sensor Development
Sony Semiconductor Solutions and Taiwan Semiconductor Manufacturing Co. announced a joint venture to develop and produce next‑generation image sensors at Sony's new plant in Kumamoto, Japan. Sony will hold a majority stake, leveraging its sensor design expertise while TSMC contributes its leading‑edge process technology and manufacturing capacity. The collaboration aims to capture emerging AI‑driven markets such as automotive imaging and robotics. Sony projects a 56% share of the global image‑sensor market by fiscal 2025, reinforcing its leadership position.
TSMC’s Sales Grow Slowest in Months Even as AI Buildout Persists
Taiwan Semiconductor Manufacturing Co. (TSMC) reported April sales up 17.5% to NT$410.7 billion (about $13.1 billion), the slowest monthly growth since October. The modest rise reflects a plateau in smartphone and consumer‑electronics demand, even as AI‑driven chip orders stay robust. Analysts expect...
OpenAI's Secret Sauce for Networking Is Now an Open Protocol. What Does It Mean for Inference?
OpenAI unveiled Multipath Reliable Connection (MRC), an open networking protocol designed to speed and stabilize data transfer between GPU clusters in massive supercomputers. The protocol, developed with AMD, Broadcom, Intel, Microsoft and NVIDIA, can link more than 130,000 GPUs using...
6.67-inch Flexible AMOLED Display Works with Raspberry Pi, LattePanda, and Other SBCs with HDMI Output
DFRobot has launched a 6.67‑inch flexible AMOLED display that ships with a dedicated MIPI‑to‑HDMI driver board, allowing plug‑and‑play operation on popular single‑board computers such as Raspberry Pi, LattePanda, Banana Pi and Orange Pi. The panel delivers a 2400 × 1080 resolution, 450 cd/m² brightness, 16.7 million colors,...
Baidu Chip Unit Kunlunxin Eyes US$14.7b Valuation in Hong Kong IPO: Sources
Kunlunxin, Baidu's AI‑chip subsidiary, is targeting a Hong Kong IPO with a valuation of at least 100 billion yuan (approximately US$14.7 billion). The unit recently launched its XPU‑R processor, aiming to power large‑language models and reduce reliance on foreign silicon. By listing in...
When Cloud Giants Meddle in Markets
Hyperscale cloud providers are buying massive volumes of DRAM and high‑bandwidth memory to fuel AI workloads, effectively hoarding a finite supply. Their pre‑emptive procurement pushes component prices higher for downstream enterprise buyers, inflating refresh costs and lengthening lead times. As...
GCT Taps Satellite Partner to Speed 5G Rollout
GCT Semiconductor has signed a reference platform agreement with a major satellite communications provider to speed the creation of 5G user equipment that works across satellite and terrestrial networks. The deal builds on an earlier chipset licensing pact and delivers...
ITC Affirms Initial Determination that Innoscience Infringed Infineon GaN Patent
The U.S. International Trade Commission affirmed its December 2025 finding that China‑based Innoscience infringed Infineon Technologies' gallium‑nitride (GaN) patent, ordering bans on imports and sales of the company's 8‑inch GaN‑on‑silicon chips. The decision is subject to a 60‑day presidential review...
Delft-Based FrostByte Secures a Cool €1.3 Million to Scale Cryogenic Electronics for Quantum Computing
Delft‑based FrostByte, a spin‑off from TU Delft and QuTech, announced a €1.3 million (≈$1.4 million) funding round led by InnovationQuarter Capital, Graduate Ventures, Paeonia Group, UNIIQ and an angel investor. The capital will fund team expansion, scale production of cryogenic switches, and...
Aixtron Supplies Planetary G5+C MOCVD Systems to Renesas
Aixtron SE has delivered multiple Planetary G5+C metal‑organic chemical vapor deposition (MOCVD) systems to Renesas Electronics, expanding the Japanese firm’s gallium nitride (GaN) production capacity for high‑volume manufacturing. The equipment, already operational, supports Renesas’ push into power‑dense applications such as...
Chip Industry Week In Review
The chip industry is seeing massive new investments, from ASE and WUS’s $1.1 billion advanced‑packaging hub in Taiwan to SpaceX’s proposed "Terafab" semiconductor complex in Texas that could total up to $119 billion. Apple is quietly discussing U.S. production of its main...
SiTime Posts 88% Revenue Growth on AI Infrastructure Demand
SiTime reported first‑quarter 2026 revenue of $113.6 million, an 88.3% year‑over‑year increase driven by soaring demand for precision timing in AI data‑center and high‑performance computing systems. While the company posted a GAAP net loss of $5.2 million, non‑GAAP earnings surged to $38.9 million,...
Infrared LEDs Support In-Cabin Sensing for Vehicle Safety
ams OSRAM introduced its OSLON Black IR:6 C‑Series infrared LEDs, a 940 nm high‑power solution designed for camera‑based in‑cabin driver and occupant monitoring. The LEDs suppress the visible red‑glow effect, delivering low‑visibility illumination that works across changing lighting conditions. By targeting...
Infineon Rad-Hard Chips Performed Flawlessly on Artemis II
NASA’s Artemis II mission completed a 10‑day crewed flight that set a new record for distance from Earth, while simultaneously proving the reliability of Infineon Technologies’ radiation‑hardened semiconductor portfolio. Infineon’s IR HiRel rad‑hard devices powered critical Orion systems, including power supply, control...
Delta Electronics Showcases Advance Semiconductor Packaging Solutions at SEMICON SEA 2026
Delta Electronics used SEMICON Southeast Asia 2026 to unveil a suite of AI‑enabled smart manufacturing solutions aimed at accelerating advanced semiconductor packaging. The showcase highlighted precision motion hardware such as the ASDA‑W3 multi‑axis servo and sub‑micron Linear Positioning Link, as...
The 100-Second Bottleneck Behind NVIDIA CPO: 7 Companies That Own the 4-Stage Test Stack
The testing stage of coherent photonic‑on‑chip (CPO) production has become the primary bottleneck, with a full optical inspection of each photonic integrated circuit taking over 100 seconds. TrendForce data and recent earnings show that seven specialist equipment firms—FormFactor, Teradyne, Keysight,...
AMD Powers Zyphra Cloud
Zyphra has partnered with AMD to power its Zyphra Cloud full‑stack AI platform using AMD Instinct MI355X GPUs on TensorWave’s purpose‑built infrastructure. The service launches with Zyphra Inference, a serverless offering that supports frontier open‑weight models such as DeepSeek V3.2,...
Intel Arc Celestial at Idle: Leak No Longer Sees Dedicated Xe3P Gaming GPUs
Recent leaks indicate Intel has scrapped plans for a dedicated Xe3P “Celestial” Arc gaming GPU line. While the Xe3P architecture remains alive in integrated graphics, mobile platforms, workstations and the Crescent Island datacenter accelerator, no discrete desktop gaming cards are...
SEMIFIVE and ICY Tech Announce 8 Nm eMRAM SoC Tape-Out at Samsung Foundry
SEMIFIVE and ICY Tech announced the successful tape‑out of an 8‑nm eMRAM‑based edge AI system‑on‑chip (SoC) using Samsung Foundry’s 8LPU‑eMRAM process. The design integrates non‑volatile magnetoresistive RAM directly on the chip, targeting low‑power, latency‑critical applications such as industrial controllers, automotive...
TSMC Signs 30‑year PPA for >1 GW Offshore Wind, Securing AI Chip Power
Taiwan Semiconductor Manufacturing Co. (TSMC) has signed a 30‑year power purchase agreement with Northland Power for the Hai Long offshore wind project, covering more than 1 GW of capacity. The deal aims to meet soaring AI‑chip power demand while bolstering Taiwan’s renewable‑energy...
Yield Redefined: From Wafer Dies to Full Lifecycle Reliability
#Technology #Thread #Semiconductor #Manufacturing #EPDT The Semiconductor EPDT Article On Yield And Modern Semiconductor: 1/ - Published In Electronic Product Design And Test (EPDT), My Latest Article Explains How Semiconductor Yield Has Expanded Beyond Traditional Wafer Metrics Into A Multi Stage Productization...
Early Nvidia AI Bet Proved 80x Success
1/ On Feb 21 I told my boss aka my wife @LTwolfe that I had my third conviction call not yet appreciated + to buy a basket *In 2016 it was NVDA, pitched publicly at Invest For Kids in Chicago (80x+)...
Iren’s Stock Soars as a Major Nvidia Investment Overshadows Revenue Shortfall
Iren’s shares surged after the company disclosed a strategic partnership with Nvidia that includes a five‑year right to purchase 30 million shares at $70 each, valuing the deal at up to $2.1 billion. The collaboration will see Iren deploy 5 gigawatts of infrastructure...
Anthropic Pledges $200 B to Google Cloud for TPU Capacity Over 5 Years
Anthropic has agreed to spend $200 billion on Google Cloud and custom TPU hardware over the next five years, a deal that represents more than 40% of Alphabet’s disclosed revenue backlog. The commitment, coupled with a $40 billion equity investment from Alphabet,...
Advent Diamond and Northrop Grumman Break 100‑W Power Record for Diamond Solid‑State Limiters
Advent Diamond Inc. and Northrop Grumman announced that a diamond‑based solid‑state limiter handled more than 100 W of RF power, setting a new performance benchmark. The achievement showcases engineered‑diamond’s readiness for high‑power defense and aerospace applications and underscores growing U.S. microelectronics...
EU‑Backed Kembara Leads $160 Million Series C in UK Quantum Motion
Kembara, the EU‑backed technology growth fund, closed a $160 million Series C round for British quantum‑computing startup Quantum Motion, marking the fund’s inaugural deployment. The round was co‑led by venture firm DCVC and included the British Business Bank and Firgun Ventures,...
Arbor ARES-2100 Wildcat Lake Fanless Box PC Targets Industrial Automation, Machine Vision, and Edge AI Applications
Arbor introduced the ARES‑2100 fanless box PC, built around Intel’s new Core Series 3 “Wildcat Lake” processor aimed at industrial automation, machine‑vision and Edge AI workloads. The compact 1U system supports up to 64 GB DDR5, UFS 3.1 or NVMe storage, three 2.5 GbE...
Nvidia Vera Rubin Used by Google Could Next and Thinking Machines Lab
NVIDIA unveiled its Vera Rubin platform, pairing the Rubin GPU with the Vera CPU to deliver a full‑stack AI system that claims ten‑fold better inference performance per watt than the Blackwell generation. At Google Cloud Next, the company announced the...
Asian Emerging‑Market Stocks Hit Record High on Iran Peace Hopes and AI Boom
Asian emerging‑market equities jumped to an all‑time high, with the MSCI Emerging Markets Index up 3.1% to about 1,700 points. The rally was driven by growing optimism over a possible US‑Iran peace settlement and a parallel AI‑driven semiconductor boom that...
Focused Helium Ions Create Ferroelectric Regions in Aluminum Nitride for Lower-Power Chips
Scientists at Oak Ridge National Laboratory have demonstrated that a tightly focused helium‑ion beam can write ferroelectric regions directly into aluminum nitride (AlN). The ion‑induced defects enable polarization switching with roughly 40% less energy and a stronger piezoresponse. Because AlN...
QNAP Unveils QAI‑h1290FX Edge AI Server for Private LLM Deployment
QNAP introduced the QAI‑h1290FX, an edge AI storage server built for on‑prem deployment of large language models. Featuring a 16‑core AMD EPYC processor, optional NVIDIA RTX GPU acceleration and a 12‑slot NVMe architecture, the system targets enterprises that need low‑latency...
RISC-V: From Niche Architecture to Strategic Foundation
At the RISC‑V Now by Andes conference, Aion Silicon’s CEO Oliver Jones declared that RISC‑V has moved from a niche alternative to a strategic foundation for modern silicon. The company’s experience delivering dozens of 7 nm and smaller designs shows the...
Integrated Voltage Regulators Help Pull Heat Out of Processors
AI accelerators in data centers are moving power conversion beneath the processor to cut parasitic losses and improve transient response. Empower Semiconductor’s Crescendo integrated voltage regulator (IVR) packs the converter, inductors and capacitors into an ultra‑thin chip, offering up to...
Nvidia Launches Spectrum‑X Ethernet Fabric and Opens MRC Protocol for Gigascale AI Clusters
Nvidia announced the Spectrum‑X AI‑native Ethernet fabric and released the Multipath Reliable Connection (MRC) transport protocol as an open specification. The move targets the massive bandwidth and resilience requirements of today’s frontier AI models, and is already deployed in leading...
AMD Beats Q1 2026 Estimates as Data Center Revenue Jumps 57% to $5.8B
Advanced Micro Devices posted a Q1 2026 earnings beat, driven by a 57% year‑over‑year rise in data‑center revenue to $5.8 billion and more than 50% growth in server CPU sales. The company forecast server‑CPU revenue to surge over 70% in the...
Culpan: Apple Goes All in on MacBook Neo Production ↦
Apple is ramping up MacBook Neo output to 10 million units, double its original forecast, after the laptop’s sales outpaced expectations and strained A18 Pro chip supplies. The company has asked TSMC for a hot‑lot of the N3E‑based chips used in...
Linux 7.2 To Support Realtek RTL8159 10GbE USB Ethernet
The Realtek RTL8159 10 Gbps USB Ethernet chipset, currently supported only by Realtek’s out‑of‑tree driver, will be merged into the mainline Linux kernel with the upcoming 7.2 release. Open‑source contributor Birger Koblitz ported the code to the r8152 driver and added...
Arm Predicts $2B AGI‑CPU Demand, Fueling AI‑driven CPU Renaissance
MyPOV: @Arm and the CPU renaissance https://t.co/Xwzt9rwrq3 Arm said the company sees $2 billion in customer demand for its AGI-CPU through 2027, double what it saw just a few weeks ago. Arm is becoming one of the headliners for what's...
Shocker SiP Currently Volatile, Will Stabilize Within a Year
Shocker SiP is lumpy... In a year or so, much less, but for now, ebbs and flows.
Kensington Boosts Charging Efficiency with New GaN Products
Kensington has broadened its USB‑C gallium‑nitride charging line with a 70W three‑port adapter, a 140W four‑port adapter, and a 240W USB‑C to USB‑C cable. The GaN‑based devices deliver high‑speed power while staying compact and cool, thanks to superior efficiency over...
AI Supply Chain Bottlenecked, Regulation Likely Biggest Constraint
$AMD and $ARM further validated this week that the entire AI supply chain is constrained. Compute, memory, storage, networking, energy, and talent. All the while regulatory may be the biggest bottleneck of all. https://t.co/mezQjhNsfi
Memory, Not Math, Drives AI's 1000× Energy Gains
To improve AI energy efficiency by 1000x, it's all about memory and data movement, not the efficiency of the arithmetic. Biomimicry of the brain, with analog in-memory compute, affords a 100x. Getting to 1000x might be unconventional: a continuous-time physical dynamical...
AMD Instinct MI350P PCIe Targets Air-Cooled Enterprise AI Servers
AMD unveiled the Instinct MI350P PCIe GPU, a dual‑slot accelerator aimed at AI inference in existing air‑cooled data‑center servers. The card packs 144 GB of HBM3E memory delivering up to 4 TB/s bandwidth and claims up to 2,299 TFLOPS (4,600 TFLOPS MXFP4) performance. It...
Google's Annual TPU Cycle Accelerates Enterprise AI Efficiency
Speed of innovation matters. Google’s move to an annual silicon cycle with the TPU 8t (training) and TPU 8i (inference) is a game-changer for enterprise AI performance and cost-efficiency. https://t.co/LRR1wOtC6I #CIO #AI #Cloud #Semiconductors #Innovation
Semiconductor Industry Faces Fragmented Standards and Runtime Assurance Gap
The semiconductor ecosystem is wrestling with fragmented standards, IP exposure, and the urgent need for runtime assurance. https://t.co/JHJJe29gFO #semiconductor #AI #AIgovernance #AIsecurity
Nota AI Signs AI Optimization Technology Supply and Strategic Partnership Agreement with Mobilint Accelerating Commercialization of On-Device AI on Domestic...
Nota AI, a specialist in AI model compression, has signed a strategic supply agreement with Korean NPU maker Mobilint to embed its NetsPresso® optimization platform into Mobilint’s MLA100 and MLA400 chips. The partnership aims to deliver low‑power, high‑performance edge AI...
TSM's Hiring Outpaces Taiwan Births, Drives US Expansion
There are 100,000 births per year in Taiwan and TSM needs to hire 10,000 people per year in Taiwan, a number which will only grow. Continued expansion in the United States is likely inevitable. Especially given immense willingness to pay a...
Arm Shares Fall Despite Strong AI CPU Demand
MyPOV: @Arm’s stock sinks even as it reveals strong interest in its CPUs for #AI servers https://t.co/UPSBFc3WSA @SiliconANGLE @Mike_Wheatley “More than four out of five of Arm’s employees are engineers, and so investors may rightly have questions about...” - @holgermu @constellationr #AIWars
AMD's Next Epyc Server Chip Debuts This Year with 256 Cores and 70% Better Performance
AMD confirmed that its next‑generation Epyc Venice processors, built on the Zen 6 architecture, will ship later this year. The chips will use TSMC’s 2 nm process, delivering up to 256 cores, a 70% performance uplift and roughly double the CPU‑to‑GPU bandwidth...
Nvidia Climbs as xAI Releases 220k Used GPUs
Nvidia up 2.6% on news that xAI is flooding the market with 220,000 secondhand GPUs. https://t.co/er8D1yd9GR